Equipment state identification method and device and electronic equipment

Abstract

The embodiment of the invention discloses an equipment state recognition method and device and electronic equipment, and the method comprises the steps: obtaining the sensor feature data of target equipment at the ith moment, and enabling the sensor feature data to at least comprise satellite positioning feature data; inputting the sensor feature data at the ith moment into a pre-trained state recognition model to obtain an initial confidence coefficient of the target equipment in at least two set states at the ith moment; inputting the initial confidence coefficient of the target equipment in the set state at the ith moment and the historical correction confidence coefficient of the target equipment in the set state before the ith moment into a pre-trained state smoothing model to obtain the correction confidence coefficient of the target equipment in the set state at the ith moment; and determining the actual state of the target equipment from the set states according to the correction confidence coefficient of the target equipment in the set states at the ith moment.

Description

Technical Field

[0001] The present disclosure relates to the field of positioning technology, and in particular to a device status identification method, apparatus, and electronic device. Background Art

[0002] At present, when the device is outdoors, GNSS positioning technology is usually used for positioning. When the device is indoors, because the GNSS signal is blocked by buildings, network positioning technology is usually used. Network positioning means that the device sends the received signals (mainly including base station signals, Wi-Fi signals, Bluetooth signals, etc.) to the network server to obtain the positioning position. Therefore, it is very important to identify whether the device is indoors or outdoors, so that the positioning service can enter the correct positioning service process. In addition, some functions based on positioning (such as check-in) also need to identify whether the device is indoors or outdoors to prevent cheating. Therefore, providing a technical solution that can accurately identify whether the device is indoors or outdoors is a problem that technicians in this field need to solve. Summary of the Invention

[0003] Embodiments of the present disclosure provide a device status identification method, apparatus, and electronic device.

[0004] In a first aspect, an embodiment of the present disclosure provides a method for identifying a device state, including:

[0005] Acquire sensor feature data of the target device at the i-th moment, wherein the sensor feature data at least includes satellite positioning feature data;

[0006] Inputting the sensor feature data at the i-th moment into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where there are at least two set states;

[0007] Inputting the initial confidence that the target device is in the set state at the i-th moment and the historical revised confidence that the target device is in the set state before the i-th moment into a pre-trained state smoothing model to obtain the revised confidence that the target device is in the set state at the i-th moment;

[0008] According to the modified confidence that the target device is in the set state at the i-th moment, the actual state of the target device is determined from the set state.

[0009] In a second aspect, an embodiment of the present disclosure provides a device state identification apparatus, comprising:

[0010] an acquisition module, which acquires sensor feature data of the target device at the i-th moment, wherein the sensor feature data at least includes satellite positioning feature data;

[0011] a pre-identification module, inputting the sensor feature data at the i-th moment into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where the set states may be at least two;

[0012] a correction module, inputting the initial confidence that the target device is in the set state at the i-th moment and the historical corrected confidence that the target device is in the set state before the i-th moment into a pre-trained state smoothing model to obtain a corrected confidence that the target device is in the set state at the i-th moment;

[0013] The determination module determines the actual state of the target device from the set state according to the modified confidence that the target device is in the set state at the i-th moment.

[0014] In a third aspect, an embodiment of the present disclosure provides an electronic device comprising a memory and at least one processor, wherein the memory is used to store one or more computer instructions, and wherein the one or more computer instructions are executed by the at least one processor to implement the above method.

[0015] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions implement the above method when executed by a processor.

[0016] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, comprising a computer program / instruction, wherein the computer program / instruction implements the above method when executed by a processor.

[0017] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects:

[0018] The above technical solution uses the sensor feature data of the target device at time i and a pre-trained state recognition model to obtain the initial confidence that the target device is in the set state at time i. Then, the initial confidence that the target device is in the set state at time i and the historical corrected confidence that the target device is in the set state before time i are input into the pre-trained state smoothing model to correct the confidence that the target device is in the set state at time i. Finally, based on the corrected confidence, the actual state of the target device can be finally determined from the set state, that is, whether the target device is actually in an indoor state, an outdoor state, or a semi-indoor state. This technical solution uses the state recognition model to determine the initial confidence that the target device is in the set state at time i, and uses the state smoothing model to correct the initial confidence that the target device is in the set state at time i to improve the accuracy of determining the confidence that the target device is in a certain set state at time i. As described above, the disclosed technical solution can not only identify which set state the target device is actually in in various scenarios, but also avoid state jumps, effectively improving the accuracy of identifying the set state of the target device.

[0019] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Other features, objectives and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0021] Figure 1 A flow chart showing a method for identifying a device state according to an embodiment of the present disclosure is shown;

[0022] Figure 2A A schematic diagram showing a plurality of satellites forming a polygon according to an embodiment of the present disclosure;

[0023] Figure 2B A schematic diagram showing the maximum azimuth angle that can be covered by a satellite according to an embodiment of the present disclosure;

[0024] Figure 3 A structural block diagram of a device state identification apparatus according to an embodiment of the present disclosure is shown;

[0025] Figure 4 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown;

[0026] Figure 5 It is a structural diagram of a computer system suitable for implementing the device status identification method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for the sake of clarity, parts not related to the description of the exemplary embodiments are omitted in the accompanying drawings.

[0028] In the present disclosure, it should be understood that terms such as "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts, or combinations thereof disclosed in the present specification, and are not intended to exclude the possibility that one or more other features, numbers, steps, actions, components, parts, or combinations thereof exist or are added.

[0029] It should also be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0030] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0031] Figure 1 A flow chart of a device status identification method according to an embodiment of the present disclosure is shown as follows: Figure 1 As shown, the device status identification method includes the following steps S101-S104:

[0032] In step S101, sensor feature data of a target device at the i-th moment is obtained, wherein the sensor feature data at least includes satellite positioning feature data;

[0033] In the present disclosure, the target device may be a mobile terminal (such as a smart phone), a wearable device (such as a smart watch), an in-vehicle multimedia system (such as a car computer), etc., which can be moved from indoors to outdoors or from outdoors to indoors. The sensor feature data refers to feature data generated based on data sensed by the sensor carried by the target device. The sensor feature data is used to identify the state of the target device. The state of the target device may include at least: being (in) indoors or being (in) outdoors. The device state identification method provided in the present disclosure can be integrated into the operating system of the target device, or integrated into the software installed on the target device.

[0034] In the present disclosure, the sensor feature data includes at least satellite positioning feature data, wherein the satellite positioning feature data is feature data generated based on satellite signals received by the positioning receiver carried by the target device. Regardless of whether the method provided by the present disclosure is integrated by an operating system or by software, the satellite positioning feature data can be generated by satellite signals received by the positioning receiver of the target device, or can be generated by software installed on the target device based on satellite signals received by the positioning receiver. The present disclosure does not impose any restrictions on this.

[0035] In step S102, the sensor feature data at the i-th moment is input into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where there are at least two set states.

[0036] Among them, at least two setting states include indoor and outdoor. Since some buildings have unenclosed or semi-enclosed corridors, in order to make the device status identification more accurate, the present invention also has three setting states, including: indoor, outdoor and semi-indoor. Among them, the semi-indoor refers to a state between indoor and outdoor. For example, if the target device is in a semi-enclosed corridor, its state is semi-indoor.

[0037] In the present disclosure, the state recognition model refers to a pre-trained model for preliminary identification of the state of the target device, whose input is the sensor feature data of the target device at the i-th moment, and the output is the initial confidence that the target device is in a certain set state at the i-th moment. When the set state includes indoor, outdoor and semi-indoor, the results output by the state recognition model include: the initial confidence that the target device is in the indoor state, the initial confidence that the target device is in the outdoor state, and the initial confidence that the target device is in the semi-indoor state. Among them, the higher the confidence value, the more reliable the recognition result of the corresponding set state, that is, the target device is most likely to be in the set state with the highest confidence.

[0038] In the present disclosure, the state recognition model can be a random forest model. Of course, those skilled in the art can use other types of models for training according to the needs of actual applications. The present disclosure does not specifically limit the specific technical selection of the state recognition model. All models that can identify the set state of the target device and output the confidence level corresponding to the preset state are within the scope of the present disclosure.

[0039] In step S103, the initial confidence that the target device is in the set state at the i-th moment and the historical revised confidence that the target device is in the set state before the i-th moment are input into a pre-trained state smoothing model to obtain the revised confidence that the target device is in the set state at the i-th moment;

[0040] In the present disclosure, the state smoothing model refers to a pre-trained model used to correct the initial confidence level of the target device being in a set state at time i, as output by the state recognition model. The model takes as input the initial confidence level of the target device being in the set state at time i and the historical corrected confidence levels of the target device being in the set state before time i. The model outputs a corrected confidence level of the target device being in the set state at time i, obtained by correcting the initial confidence level of the target device being in the set state at time i. Considering that sensors may be inaccurate due to environmental influences, resulting in errors in sensor feature data, the present disclosure uses the historical corrected confidence levels of the target device being in the set state before time i and the state smoothing model to correct the initial confidence level of the target device being in the set state at time i. This ensures the accuracy of the confidence level of the target device being in the set state at time i, thereby improving the accuracy of the recognition result of the target device being in the set state at time i and eliminating the situation where the recognition result of the target device's set state jumps due to errors in the sensor feature data at time i.

[0041] In the present disclosure, the state smoothing model can be a hidden Markov model (HMM). Those skilled in the art can use other types of models for training according to the needs of actual applications. The present disclosure does not specifically limit the specific technical selection of the state smoothing model. All models that can correct the confidence that the target device is in a set state at the i-th moment belong to the scope of the present disclosure.

[0042] In the present disclosure, the historically corrected confidence that the target device is in the set state before the i-th moment can be specifically: the historically corrected confidence that the target device is in the set state within a preset time length (also referred to as a preset time window) before the i-th moment, and the preset time length can be 10 seconds, 1 minute, etc. For example, if the i-th moment is 9:01:25 and the preset time length is 10 seconds, then the historically corrected confidence that the target device is in the set state from 9:01:15 to 24 seconds is obtained. If the set states include indoor, outdoor, and semi-indoor, the obtained historically corrected confidences are also the historically corrected confidences of each of these three states, that is, all results output by the state recognition model within the preset time length (also referred to as the preset time window) before the i-th moment are used as inputs for the i-th moment.

[0043] In step S104, the actual state of the target device is determined from the set state according to the modified confidence that the target device is in the set state at the i-th moment.

[0044] In the present disclosure, after the confidence level that the target device is in a set state at the i-th moment is corrected, the actual state of the target device can be determined from the set state based on the corrected confidence level. For example, if, after correction, the confidence level that the target device is in an indoor state at the i-th moment is higher than the confidence level that it is in an outdoor state or in a semi-outdoor state, then the actual state of the target device can be determined to be indoors; if, after correction, the confidence level that the target device is in an outdoor state at the i-th moment is higher than the confidence level that it is in an indoor state or in a semi-outdoor state, then the actual state of the target device can be determined to be outdoor; if, after correction, the confidence level that the target device is in a semi-indoor state at the i-th moment is higher than the confidence level that it is in an outdoor state or in an indoor state, then the actual state of the target device can be determined to be semi-indoor.

[0045] As described above, the present disclosure first obtains the sensor feature data of the target device at the i-th moment; then inputs the sensor feature data into a pre-trained state recognition model to pre-recognize the set state of the target device, and obtains the initial confidence that the target device is in one or several set states at the i-th moment, so as to obtain the initial recognition result for the set state of the target device at the i-th moment; then uses the pre-trained state smoothing model and the historical corrected confidence that the target device was in the set state before the i-th moment to correct the initial confidence that the target device was in one or several set states at the i-th moment, and obtains the corrected confidence that the target device was in the set state at the i-th moment. Since the historical corrected confidence that the target device was in the set state before the i-th moment represents the true state of the target device before the i-th moment, the smoothing model can improve the accuracy of determining the confidence that the target device is in the set state at the i-th moment; finally, the set state in which the target device is actually located can be determined based on the corrected confidence that the target device was in the set state at the i-th moment. The technical solution disclosed herein uses a state recognition model to determine the initial confidence that the target device is in a set state at the i-th moment, and uses a state smoothing model to correct the initial confidence that the target device is in the set state at the i-th moment, so as to improve the accuracy of determining the confidence that the target device is in a certain set state at the i-th moment. Therefore, the technical solution disclosed herein can not only identify which set state the target device is actually in under various scenarios, but also avoid state jumps, thereby effectively improving the accuracy of identifying the set state of the target device.

[0046] The above is the method provided by the embodiment of the present disclosure. The sensor characteristic data used in the present disclosure and the corresponding embodiments are introduced in detail below.

[0047] In the present disclosure, the satellite positioning feature data may include one or more of the following data:

[0048] 1. The polygonal area enclosed by satellites, including: the polygonal area of ​​visible satellites and the polygonal area of ​​theoretically visible satellites. The visible satellites refer to satellites that exist in the satellite sky map and whose Cn0 observed by the positioning receiver of the target device is not 0. The theoretically visible satellites refer to satellites that exist in the satellite sky map and whose Cn0 observed by the positioning receiver of the target device is 0. Cn0 is the carrier-to-noise power spectral density ratio of the satellite, which is often used to measure the positioning signal strength or the sensitivity of the positioning receiver. The polygonal area is the area of ​​the polygon enclosed by the position points of the satellite in the satellite sky map, such as Figure 2A As shown, if Figure 2A If all satellites shown are visible, then Figure 2A The area shown is the polygonal area of ​​visible satellites;

[0049] 2. The maximum azimuth angle of satellite coverage, including the maximum azimuth angle of visible satellites and the maximum azimuth angle of theoretically visible satellites, such as Figure 2B As shown, if Figure 2B If all satellites shown are visible, then Figure 2B Shown is the maximum azimuth of visible satellites;

[0050] 3. The number of satellites, including: 1) the number of satellites with an elevation angle greater than a preset satellite elevation angle threshold, the number of satellites including: the total number of satellites and the theoretical number of visible satellites, the maximum value of Cn0, the minimum value of Cn0, the mean value of Cn0, the standard deviation value of Cn0 of satellites with an elevation angle greater than the preset satellite elevation angle threshold, and the number of satellites whose Cn0 falls within a corresponding numerical interval, and there is at least one corresponding numerical interval; 2) the number of satellites with an elevation angle less than a preset satellite elevation angle threshold, the number of satellites including: the total number of satellites and the theoretical number of visible satellites, the maximum value of Cn0, the minimum value of Cn0, the mean value of Cn0, the standard deviation value of Cn0 of satellites with an elevation angle less than the preset satellite elevation angle threshold, and the number of satellites whose Cn0 falls within a corresponding numerical interval, and there is at least one corresponding numerical interval.

[0051] In addition, in actual applications, it may not be possible to obtain the above-mentioned satellite positioning feature data directly, but only the satellite original observation data output by the satellite original observation interface can be obtained. At this time, the positions of all satellites with elevation angles greater than 0 at the i-th moment can be calculated through the positioning position of the target device and the broadcast ephemeris, and matched with the satellites in the satellite original observation data according to the satellite system and satellite identification number. The satellites that successfully match are visible satellites, and the satellites that fail to match are theoretically visible satellites. Then, the satellite positioning feature data can be generated based on the data corresponding to the visible satellites and the theoretically visible satellites in the satellite original observation data.

[0052] As can be seen from the above, the satellite positioning feature data disclosed in the present invention not only takes into account the intensity characteristics of satellite signals, but also the distribution characteristics and visibility characteristics of satellites. For example, in a city street with tall buildings or a canyon in a mountainous area, most of the visible satellites observed by the target device are distributed in a narrow strip; in the courtyard of a building, most of the visible satellites are gathered around the center of the satellite sky map, that is, the high elevation angle area; near the window of the building or in a corridor close to the outdoors, most of the visible satellites are concentrated in a semicircle of the satellite sky map; and so on. It can be seen that the satellite positioning feature data disclosed in the present invention can characterize target devices in different scenes. Therefore, the present invention can identify the status of target devices in different scenes. That is, based on the satellite positioning feature data, the present invention can accurately identify the status of the target device in different scenes, and has stronger applicability.

[0053] Furthermore, if the target device is also equipped with sensors such as an accelerometer, a magnetometer, and a camera, then in the present disclosure, the sensor characteristic data may also include: speed characteristic data, magnetic field characteristic data, and ambient light characteristic data. It should be noted that, since the operating frequencies of different sensors are different, the present disclosure uses the time (i-th moment) of obtaining satellite positioning characteristic data as a benchmark, the speed characteristic data uses the data closest to the i-th moment, and the magnetic field characteristic data and ambient light characteristic data are the data between the i-1th moment and the i-th moment recorded and used at the i-th moment. For example, if the i-1th moment is 9:30:05 and the i-th moment is 9:30:07, then the magnetic field characteristic data and ambient light characteristic data between these two moments are recorded.

[0054] The speed characteristic data refers to characteristic data related to the speed of the target device, such as the speed value and acceleration value of the target device at a corresponding moment. The characteristic data may be provided by the operating system of the target device.

[0055] The magnetic field characteristic data refers to characteristic data related to the magnetic field in which the target device is located, such as the mean variance of the magnetic field strength detected by the magnetometer carried by the target device within a preset time period, etc.

[0056] Considering that, under the premise of normal satellite positioning, if the speed of the target device is high enough, such as above a preset speed threshold, it indicates that the target device is currently in a high-speed moving vehicle, such as a car, high-speed train, etc. Since cars and high-speed trains can only operate outdoors, at this time, no other sensor feature data is required to participate in the state recognition of the target device, and the actual state of the target device can be directly determined to be outdoor. Therefore, based on the speed feature data, it can be determined whether the magnetic field feature data needs to be used as input data for the state recognition model. In other words, the magnetic field feature data can be used to learn the state of the device's speed feature data below a preset speed threshold. For example, in the present disclosure, based on the speed feature data, it can first be determined whether the speed of the target device is below a preset speed threshold. If it is not below (above) the speed threshold, it indicates that the actual state of the target device at time i is outdoor. The target device state can then be directly output as outdoor. If it is below the speed threshold, the sensor feature data input to the state recognition model includes the magnetic field feature data. The preset speed threshold can be set according to the needs of the actual application, and the present disclosure does not specifically limit its specific value.

[0057] Among them, the ambient light characteristic data refers to characteristic data related to the ambient light of the environment in which the target device is located, such as the mean variance of the ambient light intensity detected by the target device within a preset time period, etc. Considering that the ambient light characteristic data is meaningful only in the time period with sufficient natural light. Therefore, it can be determined based on the time of the target device whether the ambient light characteristic data is used as the input data of the state recognition model, that is, the ambient light characteristic data is used for the state recognition model learning device to learn the state of the device in the time period with sufficient light. For example, in the present disclosure, it can be determined first based on the time of the target device whether the target device is in the time period with sufficient light (that is, daytime). If not, the ambient light characteristic data is not included in the sensor characteristic data input to the state recognition model. If it is, the ambient light characteristic data is included in the sensor characteristic data input to the state recognition model.

[0058] In the present disclosure, in order to further improve the recognition accuracy of state recognition, the sensor feature data input into the pre-trained state recognition model may also include map feature data and / or network signal feature data. The map feature data may include location feature data, environmental feature data, and the like of the target device, wherein the location feature data refers to feature data related to the positioning position of the device, such as the location coordinates of the positioning position, whether the positioning position is inside a building, and the like; the environmental feature data refers to feature data related to the surrounding environment of the positioning position, such as the density and average height of buildings within a preset range around the positioning position, and the like. The network signal feature data refers to feature data of the network access point scanned by the target device, and the like. The network access point feature data scanned by the target device refers to feature data related to access points such as base stations, Wifi, Bluetooth, etc. that can be scanned by the target device. The characteristic data related to the base station may include the number of base stations that the target device can scan and the signal strength of each base station, etc. The characteristic data related to the Wifi may include the number of Wifis that the target device can scan and the signal strength of each Wifi, etc. The characteristic data related to the Bluetooth may include the number of Bluetooths that the target device can scan and the signal strength of each Bluetooth, etc.

[0059] Considering that satellite signals can attenuate to varying degrees in a vehicle-mounted environment, when the speed of the vehicle carrying the target device decreases, the state recognition model, which primarily relies on satellite signals for initial identification of the target device's state, is prone to misidentifying the target device as being indoors. Therefore, the present disclosure further introduces a state transition recognition model to eliminate factors that may cause errors in the initial identification of the target device's state. Specifically, the state transition recognition model is used to adjust the state transition probability parameters of the state smoothing model. In this case, the method may also include the following steps:

[0060] From the satellite positioning feature data, ambient light feature data, and network signal feature data included in the sensor feature data acquired at and before the i-th moment, signal strength feature data representing changes in satellite signal strength, ambient light strength, and network signal strength are respectively acquired; wherein the sensor feature data acquired before the i-th moment may be specifically implemented as: sensor feature data acquired within a preset time period before the i-th moment, where the preset time period may be 10 seconds, 1 minute, etc. The preset time period here may be the same as or different from the preset time period for acquiring the historical correction confidence level;

[0061] Inputting the signal strength characteristic data into a pre-trained state transition recognition model to obtain a signal strength change recognition result of the target device;

[0062] The signal strength change identification result is input into the state smoothing model, and the value of the state transition probability of the state smoothing model is adjusted. The value of the state transition probability is used to correct the initial confidence of the target device in the set state at the i-th moment output by the state identification model.

[0063] In the present disclosure, the signal strength characteristic data refers to characteristic data related to satellite signals received by the target device, detected ambient light signals, and received network signals, used to characterize the strength of the aforementioned signals and capable of indicating a transition in the state of the target device. For example, the signal strength characteristic data may include signal strength characteristic data characterizing changes in satellite signal strength, signal strength characteristic data characterizing changes in ambient light intensity, and signal strength characteristic data characterizing changes in network signal strength, etc.

[0064] The signal strength characteristic data used to characterize the change in satellite signal strength can be extracted from the satellite positioning characteristic data included in the sensor characteristic data obtained at and before the i-th moment, and may include one or more of the following characteristic data: the total number of satellites with an elevation angle greater than a preset satellite elevation angle threshold, the number of visible satellites with an elevation angle greater than a preset satellite elevation angle threshold, the maximum and minimum values ​​of the signal strength of satellites with an elevation angle greater than the preset satellite elevation angle threshold, the average value of the signal strength of satellites with an elevation angle greater than the preset satellite elevation angle threshold, the total number of satellites with an elevation angle less than the preset satellite elevation angle threshold, the number of visible satellites with an elevation angle less than the preset satellite elevation angle threshold, the maximum and minimum values ​​of the signal strength of satellites with an elevation angle less than the preset satellite elevation angle threshold, the average value of the signal strength of satellites with an elevation angle less than the preset satellite elevation angle threshold, and the like. Of course, those skilled in the art may also use other signal strength characteristic data that can be used to characterize the change in satellite signal strength, and the present disclosure does not particularly limit the specific form of the characteristic data.

[0065] The signal intensity characteristic data used to characterize changes in ambient light intensity can be extracted from the ambient light characteristic data included in the sensor characteristic data acquired at or before time i, and can include, for example, the mean and variance of the light intensity detected by the target device. Of course, those skilled in the art may also employ other signal intensity characteristic data capable of characterizing changes in ambient light intensity, and this disclosure does not particularly limit the specific form of such characteristic data.

[0066] The signal strength characteristic data used to characterize changes in network signal strength can be extracted from the network signal characteristic data included in the sensor characteristic data acquired at and before time i, and can include, for example, the mean and variance of the network signal strength received by the target device. Of course, those skilled in the art may also use other signal strength characteristic data that can be used to characterize changes in network signal strength, and this disclosure does not particularly limit the specific form of such characteristic data.

[0067] In the present disclosure, the signal strength characteristic data may also include signal strength characteristic data for characterizing changes in magnetic field strength. The characteristic data may be extracted from the magnetic field characteristic data included in the sensor characteristic data acquired at the i-th moment and before the i-th moment. For example, it may be the mean variance of the magnetic field strength detected by the target device, and so on. Of course, those skilled in the art may also adopt other signal strength characteristic data that can be used to characterize changes in magnetic field strength. The present disclosure does not specifically limit the specific form of expression of the characteristic data. When the signal strength characteristic data includes the signal strength characteristic data for characterizing changes in magnetic field strength, the signal strength characteristic data for characterizing changes in magnetic field strength may be input into a pre-trained state transition recognition model together with the above-mentioned signal strength characteristic data for characterizing changes in satellite signal strength, changes in ambient light intensity, and changes in network signal strength.

[0068] In the present disclosure, the state transition recognition model refers to a pre-trained model used to identify changes in features that may cause a transition in the recognition result of the target device's state. The state transition recognition model can be a decision tree model, whose input is the signal strength feature data and whose output is the recognition result of changes in the signal strength of the satellite signal, the detected ambient light signal, and the received network signal received by the target device. Based on the above signal strength change recognition result, it can be determined whether a signal is significantly enhanced, unchanged, or significantly weakened. By using the recognition result of the state transition recognition model and adjusting the state transition probability value in the state smoothing model based on the recognition result of the state transition recognition model, the initial confidence level that the target device is in the set state at time i is corrected, ultimately achieving the purpose of improving the accuracy of the confidence level that the target device is in the set state at time i, thereby improving the accuracy of the recognition result of the target device's set state and reducing the error in the recognition result of the target device's set state.

[0069] When the set states include indoor, outdoor and semi-indoor, the state transition probability values ​​of the state smoothing model may include: the probability value of indoor to outdoor, the probability value of indoor to semi-outdoor, the probability value of indoor to indoor; the probability value of outdoor to indoor, the probability value of outdoor to semi-indoor, the probability value of outdoor to outdoor; the probability value of semi-indoor to outdoor, the probability value of semi-indoor to indoor, the probability value of semi-indoor to semi-indoor. When the signal strength change identification result at least indicates that the satellite signal received by the target device is significantly enhanced, or the ambient light is significantly enhanced, it means that the target device is very likely to be moving from indoors or semi-indoor to outdoors. At this time, the state transition probability values ​​of indoor to outdoor and semi-indoor to outdoor in the state smoothing model can be increased respectively. The state transition probability value of the state smoothing model remains unchanged. When the signal strength change identification result at least indicates that the network signal received by the target device is significantly enhanced, the satellite signal is significantly weakened, or the ambient light is significantly weakened, it means that the target device is very likely to be moving from outdoor or semi-indoor to indoor. At this time, the state transition probability values ​​of the state smoothing model from outdoor to indoor and from semi-indoor to indoor can be increased respectively, and the other state transition probability values ​​in the state smoothing model remain unchanged. When the signal strength change identification result at least indicates that the signal strength received by the target device remains unchanged, the state transition probability value in the state smoothing model may not be adjusted, and the state transition probability value in the state smoothing model remains unchanged.

[0070] That is, in the present disclosure, the step of adjusting the value of the state transition probability of the state smoothing model may include the following steps:

[0071] When the signal strength change identification result at least indicates that the satellite signal received by the target device is significantly enhanced or the ambient light is significantly enhanced, the state transition probability values ​​from indoor to outdoor and from semi-indoor to outdoor in the state smoothing model are increased respectively;

[0072] When the signal strength change identification result at least indicates that the network signal received by the target device is significantly enhanced, the satellite signal is significantly weakened, or the ambient light is significantly weakened, the state transition probability values ​​from outdoor to indoor and from semi-indoor to indoor in the state smoothing model are increased respectively.

[0073] As mentioned above, the state recognition model of the present disclosure is obtained by pre-training a model for performing preliminary recognition of the state of the target device. Therefore, the method provided by the present disclosure may further include the following steps:

[0074] Acquire training data for a state recognition model, the training data including sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data;

[0075] Determine a status tag of a device corresponding to the sensor characteristic data;

[0076] The sensor feature data and the state label of the corresponding device are used as training input data of the state recognition model to train the state recognition model.

[0077] In this embodiment, when training the state recognition model, training data of the state recognition model including sensor feature data of the device is first obtained, wherein the sensor feature data includes at least satellite positioning feature data, and the sensor feature data may be received by the device at a certain historical moment or received by the device within a certain historical time period. Then, a state label of the device corresponding to the sensor feature data is determined. When the sensor feature data is sensor feature data received by the device at a certain historical moment, the state label of the device corresponding to the sensor feature data is whether the set state of the device at that historical moment is indoors, outdoor, or semi-indoor. When the sensor feature data is sensor feature data received by the device within a certain historical time period, the state label of the device corresponding to the sensor feature data is whether the set state of the device at each historical moment within the historical time period is indoors, outdoor, or semi-indoor. Finally, the sensor feature data and the corresponding state label of the device are used as training input data of the state recognition model to train the state recognition model. When the state recognition model meets a preset convergence condition, the training is terminated to obtain the state recognition model.

[0078] As mentioned above, when the state recognition model is used to identify the set state of the target device, the sensor feature data may also include speed feature data, magnetic field feature data and ambient light feature data. Therefore, when the state recognition model is trained, the sensor feature data serving as the training input data of the state recognition model may also include speed feature data, magnetic field feature data and ambient light feature data.

[0079] As mentioned above, if the target device's speed is sufficiently high, such as exceeding a preset speed threshold, assuming normal satellite positioning, it indicates that the target device is currently in a high-speed moving vehicle, and the target device's actual location can be directly determined to be outdoors. Therefore, the speed characteristic data can be used to determine whether the magnetic field characteristic data should be used as input data for the state recognition model to initially identify the target device as being in the set state at time i. As mentioned above, the ambient light characteristic data is only meaningful during periods of sufficient natural light. Therefore, the target device's time can be used to determine whether the ambient light characteristic data should be used as input data for the state recognition model to initially identify the target device as being in the set state at time i. The same principle applies when training the state recognition model: The speed characteristic data can be used to determine whether the magnetic field characteristic data should be used as training input data for the state recognition model, and the device's time can be used to determine whether the ambient light characteristic data should be used as training input data for the state recognition model.

[0080] That is, in the present disclosure, the method further comprises:

[0081] If the speed characteristic data is lower than a preset speed threshold, the sensor characteristic data further includes: magnetic field characteristic data;

[0082] If it is determined that the device is in a time period with sufficient lighting based on the time of the device corresponding to the sensor characteristic data, the sensor characteristic data further includes: ambient light characteristic data.

[0083] As mentioned above, to further improve state recognition accuracy, the sensor feature data input into the state recognition model for preliminary identification of the target device's set state at time i may also include map feature data for the target device. When the sensor feature data input into the state recognition model also includes map feature data, when training the state recognition model, the sensor feature data used as training input for the state recognition model must also include map feature data for the device corresponding to the sensor feature data.

[0084] In the present disclosure, the method may further include the following steps:

[0085] Acquire training data for a state transition model, the training data including sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data, ambient light feature data, and network signal feature data;

[0086] Acquire signal strength characteristic data representing changes in satellite signal strength, changes in ambient light intensity, and changes in network signal strength from the sensor characteristic data;

[0087] determining a signal strength change tag corresponding to the device and the signal strength characteristic data;

[0088] The signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data are used as training input data of the state transition recognition model to train the state transition recognition model.

[0089] In this embodiment, when training the state transition recognition model, training data of the state transition model including sensor feature data of a device is first obtained, wherein the sensor feature data includes at least satellite positioning feature data, ambient light feature data, and network signal feature data; then, signal strength feature data for characterizing changes in satellite signal strength, ambient light intensity, and network signal strength are respectively obtained from the sensor feature data; then, a signal strength change label corresponding to the device and the signal strength feature data is determined. For example, if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the device and the signal strength feature data is "significant satellite signal enhancement"; if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is significantly weakened, then the signal strength change label corresponding to the device and the signal strength feature data is "significant satellite signal weakening"; if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is unchanged, then the signal strength change label corresponding to the device and the signal strength feature data is "satellite signal unchanged", and so on; if it is determined according to the signal strength feature data that the ambient light detected by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the device and the signal strength feature data is "ambient light enhancement" Obviously, if it is determined according to the signal strength characteristic data that the ambient light detected by the corresponding device is significantly weakened, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the ambient light is significantly weakened; if it is determined according to the signal strength characteristic data that the ambient light detected by the corresponding device is unchanged, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the ambient light is unchanged, and so on; if it is determined according to the signal strength characteristic data that the network signal received by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the network signal is significantly enhanced; if it is determined according to the signal strength characteristic data that the network signal received by the corresponding device is significantly enhanced If the network signal is significantly weakened, the signal strength change label corresponding to the device and the signal strength characteristic data is that the network signal is significantly weakened. If it is determined based on the signal strength characteristic data that the network signal received by the corresponding device is unchanged, then the signal strength change label corresponding to the device and the signal strength characteristic data is that the network signal is unchanged, and so on. Finally, the signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data are used as training input data for the state transition recognition model to train the state transition recognition model. When the state transition recognition model meets the preset convergence condition, the training can be terminated to obtain the state transition recognition model.

[0090] In the present disclosure, in order to further improve the recognition accuracy of the state transition recognition model, further sliding window processing may be performed on the signal strength feature data.

[0091] That is, in the present disclosure, the method may further include the following steps:

[0092] Based on the time of the signal strength characteristic data, according to a preset time length, counting the characteristic data extreme values ​​of the signal strength characteristic data in different time sliding windows;

[0093] Calculating a characteristic data difference between the characteristic data extreme value and the value of the signal strength characteristic data at the current moment;

[0094] The using the signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data as training input data of the state transition recognition model to train the state transition recognition model includes:

[0095] The characteristic data difference and the signal strength change label corresponding to the characteristic data difference are used as training input data of the state transition recognition model to train the state transition recognition model.

[0096] In this embodiment, first, based on the time of the signal strength characteristic data, according to the preset time length, the characteristic data extreme values ​​of the signal strength characteristic data in different time sliding windows are counted, wherein the preset time length can be set according to the needs of the actual application, for example, it can be set to 5s or other values, wherein for visible satellites, the maximum value of the corresponding characteristic data can be counted, and for theoretically visible satellites, the minimum value of the corresponding characteristic data can be counted; then, the characteristic data difference between the statistically obtained characteristic data extreme value and the value of the signal strength characteristic data at the current moment is calculated; finally, the characteristic data difference is used to replace the signal strength characteristic data received by the training device to train the state transition recognition model, that is, the characteristic data difference and the signal strength change label corresponding to the characteristic data difference are used as the training input data of the state transition recognition model, and the state transition recognition model is trained. When the state transition recognition model meets the preset convergence condition, the training can be ended to obtain the state transition recognition model.

[0097] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein.

[0098] Figure 3 The following is a structural block diagram of a device state recognition apparatus according to an embodiment of the present disclosure. The apparatus can be implemented as part or all of an electronic device through software, hardware, or a combination of both. Figure 3 As shown, the device status identification device includes:

[0099] An acquisition module 301 acquires sensor feature data of a target device at the i-th moment, wherein the sensor feature data includes at least satellite positioning feature data;

[0100] In the present disclosure, the target device may be a mobile terminal (such as a smart phone), a wearable device (such as a smart watch), an in-vehicle multimedia system (such as a car computer), etc., which can be moved from indoors to outdoors or from outdoors to indoors. The sensor feature data refers to feature data generated based on data sensed by the sensor carried by the target device. The sensor feature data is used to identify the state of the target device. The state of the target device may at least include: being (in) indoors or being (in) outdoors. The device state identification device provided in the present disclosure may be implemented as an operating system of the target device, or as software installed on the target device.

[0101] In the present disclosure, the sensor feature data includes at least satellite positioning feature data, wherein the satellite positioning feature data is feature data generated based on satellite signals received by the positioning receiver carried by the target device. Regardless of whether the device provided by the present disclosure is implemented as an operating system or software, the satellite positioning feature data can be generated by the satellite signals received by the positioning receiver of the target device, or can be generated by software installed on the target device based on the satellite signals received by the positioning receiver. The present disclosure does not impose any restrictions on this.

[0102] A pre-identification module 302 inputs the sensor feature data at the i-th moment into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where there are at least two set states.

[0103] Among them, at least two setting states include indoor and outdoor. Since some buildings have unenclosed or semi-enclosed corridors, in order to make the device status identification more accurate, the present invention also has three setting states, including: indoor, outdoor and semi-indoor. Among them, the semi-indoor refers to a state between indoor and outdoor. For example, if the target device is in a semi-enclosed corridor, its state is semi-indoor.

[0104] In the present disclosure, the state recognition model refers to a pre-trained model for preliminary identification of the state of the target device, whose input is the sensor feature data of the target device at the i-th moment, and the output is the initial confidence that the target device is in a certain set state at the i-th moment. When the set state includes indoor, outdoor and semi-indoor, the results output by the state recognition model include: the initial confidence that the target device is in the indoor state, the initial confidence that the target device is in the outdoor state, and the initial confidence that the target device is in the semi-indoor state. Among them, the higher the confidence value, the more reliable the recognition result of the corresponding set state, that is, the target device is most likely to be in the set state with the highest confidence.

[0105] In the present disclosure, the state recognition model can be a random forest model. Of course, those skilled in the art can use other types of models for training according to the needs of actual applications. The present disclosure does not specifically limit the specific technical selection of the state recognition model. All models that can identify the set state of the target device and output the confidence level corresponding to the preset state are within the scope of the present disclosure.

[0106] The correction module 303 inputs the initial confidence that the target device is in the set state at the i-th moment and the historical corrected confidence that the target device is in the set state before the i-th moment into a pre-trained state smoothing model to obtain a corrected confidence that the target device is in the set state at the i-th moment;

[0107] In the present disclosure, the state smoothing model refers to a pre-trained model used to correct the initial confidence level of the target device being in a set state at time i, as output by the state recognition model. The model takes as input the initial confidence level of the target device being in the set state at time i and the historical corrected confidence levels of the target device being in the set state before time i. The model outputs a corrected confidence level of the target device being in the set state at time i, obtained by correcting the initial confidence level of the target device being in the set state at time i. Considering that sensors may be inaccurate due to environmental influences, resulting in errors in sensor feature data, the present disclosure uses the historical corrected confidence levels of the target device being in the set state before time i and the state smoothing model to correct the initial confidence level of the target device being in the set state at time i. This ensures the accuracy of the confidence level of the target device being in the set state at time i, thereby improving the accuracy of the recognition result of the target device being in the set state at time i and eliminating the situation where the recognition result of the target device's set state jumps due to errors in the sensor feature data at time i.

[0108] In the present disclosure, the state smoothing model can be a hidden Markov model (HMM). Those skilled in the art can use other types of models for training according to the needs of actual applications. The present disclosure does not specifically limit the specific technical selection of the state smoothing model. All models that can correct the confidence that the target device is in a set state at the i-th moment belong to the scope of the present disclosure.

[0109] In the present disclosure, the historically corrected confidence that the target device is in the set state before the i-th moment can be specifically: the historically corrected confidence that the target device is in the set state within a preset time length (also referred to as a preset time window) before the i-th moment, and the preset time length can be 10 seconds, 1 minute, etc. For example, if the i-th moment is 9:01:25 and the preset time length is 10 seconds, then the historically corrected confidence that the target device is in the set state from 9:01:15 to 24 seconds is obtained. If the set states include indoor, outdoor, and semi-indoor, the obtained historically corrected confidences are also the historically corrected confidences of each of these three states, that is, all results output by the state recognition model within the preset time length (also referred to as the preset time window) before the i-th moment are used as inputs for the i-th moment.

[0110] The determination module 304 determines the actual state of the target device from the set state according to the modified confidence that the target device is in the set state at the i-th moment.

[0111] In the present disclosure, after the confidence level that the target device is in a set state at the i-th moment is corrected, the actual state of the target device can be determined from the set state based on the corrected confidence level. For example, if, after correction, the confidence level that the target device is in an indoor state at the i-th moment is higher than the confidence level that it is in an outdoor state or in a semi-outdoor state, then the actual state of the target device can be determined to be indoors; if, after correction, the confidence level that the target device is in an outdoor state at the i-th moment is higher than the confidence level that it is in an indoor state or in a semi-outdoor state, then the actual state of the target device can be determined to be outdoor; if, after correction, the confidence level that the target device is in a semi-indoor state at the i-th moment is higher than the confidence level that it is in an outdoor state or in an indoor state, then the actual state of the target device can be determined to be semi-indoor.

[0112] As described above, the present disclosure first obtains the sensor feature data of the target device at the i-th moment; then inputs the sensor feature data into a pre-trained state recognition model to pre-recognize the set state of the target device, and obtains the initial confidence that the target device is in one or several set states at the i-th moment, so as to obtain the initial recognition result for the set state of the target device at the i-th moment; then uses the pre-trained state smoothing model and the historical corrected confidence that the target device was in the set state before the i-th moment to correct the initial confidence that the target device was in one or several set states at the i-th moment, and obtains the corrected confidence that the target device was in the set state at the i-th moment. Since the historical corrected confidence that the target device was in the set state before the i-th moment represents the true state of the target device before the i-th moment, the smoothing model can improve the accuracy of determining the confidence that the target device is in the set state at the i-th moment; finally, the set state in which the target device is actually located can be determined based on the corrected confidence that the target device was in the set state at the i-th moment. The technical solution disclosed herein uses a state recognition model to determine the initial confidence that the target device is in a set state at the i-th moment, and uses a state smoothing model to correct the initial confidence that the target device is in the set state at the i-th moment, so as to improve the accuracy of determining the confidence that the target device is in a certain set state at the i-th moment. Therefore, the technical solution disclosed herein can not only identify which set state the target device is actually in under various scenarios, but also avoid state jumps, thereby effectively improving the accuracy of identifying the set state of the target device.

[0113] The above is the device provided by the embodiment of the present disclosure. The sensor characteristic data used in the present disclosure and the corresponding embodiments are introduced in detail below.

[0114] In the present disclosure, the satellite positioning feature data may include one or more of the following data:

[0115] 1. The polygonal area enclosed by satellites, including: the polygonal area of ​​visible satellites and the polygonal area of ​​theoretically visible satellites. The visible satellites refer to satellites that exist in the satellite sky map and whose Cn0 observed by the positioning receiver of the target device is not 0. The theoretically visible satellites refer to satellites that exist in the satellite sky map and whose Cn0 observed by the positioning receiver of the target device is 0. Cn0 is the carrier-to-noise power spectral density ratio of the satellite, which is often used to measure the positioning signal strength or the sensitivity of the positioning receiver. The polygonal area is the area of ​​the polygon enclosed by the position points of the satellite in the satellite sky map, such as Figure 2A As shown, if Figure 2A If all satellites shown are visible, then Figure 2A The area shown is the polygonal area of ​​visible satellites;

[0116] 2. The maximum azimuth angle of satellite coverage, including the maximum azimuth angle of visible satellites and the maximum azimuth angle of theoretically visible satellites, such as Figure 2B As shown, if Figure 2B If all satellites shown are visible, then Figure 2B Shown is the maximum azimuth of visible satellites;

[0117] 3. The number of satellites, including: 1) the number of satellites with an elevation angle greater than a preset satellite elevation angle threshold, the number of satellites including: the total number of satellites and the theoretical number of visible satellites, the maximum value of Cn0, the minimum value of Cn0, the mean value of Cn0, the standard deviation value of Cn0 of satellites with an elevation angle greater than the preset satellite elevation angle threshold, and the number of satellites whose Cn0 falls within a corresponding numerical interval, and there is at least one corresponding numerical interval; 2) the number of satellites with an elevation angle less than a preset satellite elevation angle threshold, the number of satellites including: the total number of satellites and the theoretical number of visible satellites, the maximum value of Cn0, the minimum value of Cn0, the mean value of Cn0, the standard deviation value of Cn0 of satellites with an elevation angle less than the preset satellite elevation angle threshold, and the number of satellites whose Cn0 falls within a corresponding numerical interval, and there is at least one corresponding numerical interval.

[0118] In addition, in actual applications, it may not be possible to obtain the above-mentioned satellite positioning feature data directly, but only the satellite original observation data output by the satellite original observation interface can be obtained. At this time, the positions of all satellites with elevation angles greater than 0 at the i-th moment can be calculated through the positioning position of the target device and the broadcast ephemeris, and matched with the satellites in the satellite original observation data according to the satellite system and satellite identification number. The satellites that successfully match are visible satellites, and the satellites that fail to match are theoretically visible satellites. Then, the satellite positioning feature data can be generated based on the data corresponding to the visible satellites and the theoretically visible satellites in the satellite original observation data.

[0119] As can be seen from the above, the satellite positioning feature data disclosed in the present invention not only takes into account the intensity characteristics of satellite signals, but also the distribution characteristics and visibility characteristics of satellites. For example, in a city street with tall buildings or a canyon in a mountainous area, most of the visible satellites observed by the target device are distributed in a narrow strip; in the courtyard of a building, most of the visible satellites are gathered around the center of the satellite sky map, that is, the high elevation angle area; near the window of the building or in a corridor close to the outdoors, most of the visible satellites are concentrated in a semicircle of the satellite sky map; and so on. It can be seen that the satellite positioning feature data disclosed in the present invention can characterize target devices in different scenes. Therefore, the present invention can identify the status of target devices in different scenes. That is, based on the satellite positioning feature data, the present invention can accurately identify the status of the target device in different scenes, and has stronger applicability.

[0120] Furthermore, if the target device is also equipped with sensors such as an accelerometer, a magnetometer, and a camera, then in the present disclosure, the sensor characteristic data may also include: speed characteristic data, magnetic field characteristic data, and ambient light characteristic data. It should be noted that, since the operating frequencies of different sensors are different, the present disclosure uses the time (i-th moment) of obtaining satellite positioning characteristic data as a benchmark, the speed characteristic data uses the data closest to the i-th moment, and the magnetic field characteristic data and ambient light characteristic data are the data between the i-1th moment and the i-th moment recorded and used at the i-th moment. For example, if the i-1th moment is 9:30:05 and the i-th moment is 9:30:07, then the magnetic field characteristic data and ambient light characteristic data between these two moments are recorded.

[0121] The speed characteristic data refers to characteristic data related to the speed of the target device, such as the speed value and acceleration value of the target device at a corresponding moment. The characteristic data may be provided by the operating system of the target device.

[0122] The magnetic field characteristic data refers to characteristic data related to the magnetic field in which the target device is located, such as the mean variance of the magnetic field strength detected by the magnetometer carried by the target device within a preset time period, etc.

[0123] Considering that, under the premise of normal satellite positioning, if the speed of the target device is high enough, such as above a preset speed threshold, it indicates that the target device is currently in a high-speed moving vehicle, such as a car, high-speed train, etc. Since cars and high-speed trains can only operate outdoors, at this time, no other sensor feature data is required to participate in the state recognition of the target device, and the actual state of the target device can be directly determined to be outdoor. Therefore, based on the speed feature data, it can be determined whether the magnetic field feature data needs to be used as input data for the state recognition model. In other words, the magnetic field feature data can be used to learn the state of the device's speed feature data below a preset speed threshold. For example, in the present disclosure, based on the speed feature data, it can first be determined whether the speed of the target device is below a preset speed threshold. If it is not below (above) the speed threshold, it indicates that the actual state of the target device at time i is outdoor. The target device state can then be directly output as outdoor. If it is below the speed threshold, the sensor feature data input to the state recognition model includes the magnetic field feature data. The preset speed threshold can be set according to the needs of the actual application, and the present disclosure does not specifically limit its specific value.

[0124] Among them, the ambient light characteristic data refers to characteristic data related to the ambient light of the environment in which the target device is located, such as the mean variance of the ambient light intensity detected by the target device within a preset time period, etc. Considering that the ambient light characteristic data is meaningful only in the time period with sufficient natural light. Therefore, it can be determined based on the time of the target device whether the ambient light characteristic data is used as the input data of the state recognition model, that is, the ambient light characteristic data is used for the state recognition model learning device to learn the state of the device in the time period with sufficient light. For example, in the present disclosure, it can be determined first based on the time of the target device whether the target device is in the time period with sufficient light (that is, daytime). If not, the ambient light characteristic data is not included in the sensor characteristic data input to the state recognition model. If it is, the ambient light characteristic data is included in the sensor characteristic data input to the state recognition model.

[0125] In the present disclosure, in order to further improve the recognition accuracy of state recognition, the sensor feature data input into the pre-trained state recognition model may also include map feature data and / or network signal feature data. The map feature data may include location feature data, environmental feature data, and the like of the target device, wherein the location feature data refers to feature data related to the positioning position of the device, such as the location coordinates of the positioning position, whether the positioning position is inside a building, and the like; the environmental feature data refers to feature data related to the surrounding environment of the positioning position, such as the density and average height of buildings within a preset range around the positioning position, and the like. The network signal feature data refers to feature data of the network access point scanned by the target device, and the like. The network access point feature data scanned by the target device refers to feature data related to access points such as base stations, Wifi, Bluetooth, etc. that can be scanned by the target device. The characteristic data related to the base station may include the number of base stations that the target device can scan and the signal strength of each base station, etc. The characteristic data related to the Wifi may include the number of Wifis that the target device can scan and the signal strength of each Wifi, etc. The characteristic data related to the Bluetooth may include the number of Bluetooths that the target device can scan and the signal strength of each Bluetooth, etc.

[0126] Considering that satellite signals may attenuate to varying degrees in a vehicle-mounted environment, when the speed of the vehicle carrying the target device decreases, the state recognition model that primarily relies on satellite signals for initial identification of the target device's state is prone to misidentifying the target device as being indoors. Therefore, the present disclosure further introduces a state transition recognition model to eliminate factors that may cause errors in the initial identification of the target device's state. Specifically, the state transition recognition model is used to adjust the state transition probability parameters of the state smoothing model. In this case, the apparatus may also include:

[0127] The state transition identification module obtains signal strength characteristic data representing changes in satellite signal strength, ambient light intensity, and network signal strength, respectively, from the satellite positioning characteristic data, ambient light characteristic data, and network signal characteristic data, including the sensor characteristic data obtained at and before the i-th moment. The sensor characteristic data obtained before the i-th moment may be specifically implemented as sensor characteristic data obtained within a preset time period before the i-th moment, where the preset time period may be 10 seconds, 1 minute, or the like. The preset time period here may be the same as or different from the preset time period for obtaining the historically corrected confidence level. The signal strength characteristic data is input into a pre-trained state transition identification model to obtain a signal strength change identification result for the target device. The signal strength change identification result is input into the state smoothing model to adjust the state transition probability value of the state smoothing model. The state transition probability value is used to correct the initial confidence level output by the state identification model that the target device is in the set state at the i-th moment.

[0128] In the present disclosure, the signal strength characteristic data refers to characteristic data related to satellite signals received by the target device, detected ambient light signals, and received network signals, used to characterize the strength of the aforementioned signals and capable of indicating a transition in the state of the target device. For example, the signal strength characteristic data may include signal strength characteristic data characterizing changes in satellite signal strength, signal strength characteristic data characterizing changes in ambient light intensity, and signal strength characteristic data characterizing changes in network signal strength, etc.

[0129] The signal strength characteristic data used to characterize the change in satellite signal strength can be extracted from the satellite positioning characteristic data included in the sensor characteristic data obtained at and before the i-th moment, and may include one or more of the following characteristic data: the total number of satellites with an elevation angle greater than a preset satellite elevation angle threshold, the number of visible satellites with an elevation angle greater than a preset satellite elevation angle threshold, the maximum and minimum values ​​of the signal strength of satellites with an elevation angle greater than the preset satellite elevation angle threshold, the average value of the signal strength of satellites with an elevation angle greater than the preset satellite elevation angle threshold, the total number of satellites with an elevation angle less than the preset satellite elevation angle threshold, the number of visible satellites with an elevation angle less than the preset satellite elevation angle threshold, the maximum and minimum values ​​of the signal strength of satellites with an elevation angle less than the preset satellite elevation angle threshold, the average value of the signal strength of satellites with an elevation angle less than the preset satellite elevation angle threshold, and the like. Of course, those skilled in the art may also use other signal strength characteristic data that can be used to characterize the change in satellite signal strength, and the present disclosure does not particularly limit the specific form of the characteristic data.

[0130] The signal intensity characteristic data used to characterize changes in ambient light intensity can be extracted from the ambient light characteristic data included in the sensor characteristic data acquired at or before time i, and can include, for example, the mean and variance of the light intensity detected by the target device. Of course, those skilled in the art may also employ other signal intensity characteristic data capable of characterizing changes in ambient light intensity, and this disclosure does not particularly limit the specific form of such characteristic data.

[0131] The signal strength characteristic data used to characterize changes in network signal strength can be extracted from the network signal characteristic data included in the sensor characteristic data acquired at and before time i, and can include, for example, the mean and variance of the network signal strength received by the target device. Of course, those skilled in the art may also use other signal strength characteristic data that can be used to characterize changes in network signal strength, and this disclosure does not particularly limit the specific form of such characteristic data.

[0132] In the present disclosure, the signal strength characteristic data may also include signal strength characteristic data for characterizing changes in magnetic field strength. The characteristic data may be extracted from the magnetic field characteristic data included in the sensor characteristic data acquired at the i-th moment and before the i-th moment. For example, it may be the mean variance of the magnetic field strength detected by the target device, and so on. Of course, those skilled in the art may also adopt other signal strength characteristic data that can be used to characterize changes in magnetic field strength. The present disclosure does not specifically limit the specific form of expression of the characteristic data. When the signal strength characteristic data includes the signal strength characteristic data for characterizing changes in magnetic field strength, the signal strength characteristic data for characterizing changes in magnetic field strength may be input into a pre-trained state transition recognition model together with the above-mentioned signal strength characteristic data for characterizing changes in satellite signal strength, changes in ambient light intensity, and changes in network signal strength.

[0133] In the present disclosure, the state transition recognition model refers to a pre-trained model used to identify changes in features that may cause a transition in the recognition result of the target device's state. The state transition recognition model can be a decision tree model, whose input is the signal strength feature data and whose output is the recognition result of changes in the signal strength of the satellite signal, the detected ambient light signal, and the received network signal received by the target device. Based on the above signal strength change recognition result, it can be determined whether a signal is significantly enhanced, unchanged, or significantly weakened. By using the recognition result of the state transition recognition model and adjusting the state transition probability value in the state smoothing model based on the recognition result of the state transition recognition model, the initial confidence level that the target device is in the set state at time i is corrected, ultimately achieving the purpose of improving the accuracy of the confidence level that the target device is in the set state at time i, thereby improving the accuracy of the recognition result of the target device's set state and reducing the error in the recognition result of the target device's set state.

[0134] When the set states include indoor, outdoor and semi-indoor, the state transition probability values ​​of the state smoothing model may include: the probability value of indoor to outdoor, the probability value of indoor to semi-outdoor, the probability value of indoor to indoor; the probability value of outdoor to indoor, the probability value of outdoor to semi-indoor, the probability value of outdoor to outdoor; the probability value of semi-indoor to outdoor, the probability value of semi-indoor to indoor, the probability value of semi-indoor to semi-indoor. When the signal strength change identification result at least indicates that the satellite signal received by the target device is significantly enhanced, or the ambient light is significantly enhanced, it means that the target device is very likely to be moving from indoors or semi-indoor to outdoors. At this time, the state transition probability values ​​of indoor to outdoor and semi-indoor to outdoor in the state smoothing model can be increased respectively. The state transition probability value of the state smoothing model remains unchanged. When the signal strength change identification result at least indicates that the network signal received by the target device is significantly enhanced, the satellite signal is significantly weakened, or the ambient light is significantly weakened, it means that the target device is very likely to be moving from outdoor or semi-indoor to indoor. At this time, the state transition probability values ​​of the state smoothing model from outdoor to indoor and from semi-indoor to indoor can be increased respectively, and the other state transition probability values ​​in the state smoothing model remain unchanged. When the signal strength change identification result at least indicates that the signal strength received by the target device remains unchanged, the state transition probability value in the state smoothing model may not be adjusted, and the state transition probability value in the state smoothing model remains unchanged.

[0135] That is, in the present disclosure, the portion for adjusting the value of the state transition probability of the state smoothing model may be configured as follows:

[0136] When the signal strength change identification result at least indicates that the satellite signal received by the target device is significantly enhanced or the ambient light is significantly enhanced, the state transition probability values ​​from indoor to outdoor and from semi-indoor to outdoor in the state smoothing model are increased respectively;

[0137] When the signal strength change identification result at least indicates that the network signal received by the target device is significantly enhanced, the satellite signal is significantly weakened, or the ambient light is significantly weakened, the state transition probability values ​​from outdoor to indoor and from semi-indoor to indoor in the state smoothing model are increased respectively.

[0138] As mentioned above, the state recognition model of the present disclosure is obtained by pre-training a model for performing preliminary recognition of the state of the target device. Therefore, the apparatus provided by the present disclosure may further include:

[0139] The first training module obtains training data for the state recognition model, where the training data includes sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data; determines the state label of the device corresponding to the sensor feature data; and uses the sensor feature data and the state label of the corresponding device as training input data for the state recognition model to train the state recognition model.

[0140] In this embodiment, when training the state recognition model, training data of the state recognition model including sensor feature data of the device is first obtained, wherein the sensor feature data includes at least satellite positioning feature data, and the sensor feature data may be received by the device at a certain historical moment or received by the device within a certain historical time period. Then, a state label of the device corresponding to the sensor feature data is determined. When the sensor feature data is sensor feature data received by the device at a certain historical moment, the state label of the device corresponding to the sensor feature data is whether the set state of the device at that historical moment is indoors, outdoor, or semi-indoor. When the sensor feature data is sensor feature data received by the device within a certain historical time period, the state label of the device corresponding to the sensor feature data is whether the set state of the device at each historical moment within the historical time period is indoors, outdoor, or semi-indoor. Finally, the sensor feature data and the corresponding state label of the device are used as training input data of the state recognition model to train the state recognition model. When the state recognition model meets a preset convergence condition, the training is terminated to obtain the state recognition model.

[0141] As mentioned above, when the state recognition model is used to identify the set state of the target device, the sensor feature data may also include speed feature data, magnetic field feature data and ambient light feature data. Therefore, when the state recognition model is trained, the sensor feature data serving as the training input data of the state recognition model may also include speed feature data, magnetic field feature data and ambient light feature data.

[0142] As mentioned above, if the target device's speed is sufficiently high, such as exceeding a preset speed threshold, assuming normal satellite positioning, it indicates that the target device is currently in a high-speed moving vehicle, and the target device's actual location can be directly determined to be outdoors. Therefore, the speed characteristic data can be used to determine whether the magnetic field characteristic data should be used as input data for the state recognition model to initially identify the target device as being in the set state at time i. As mentioned above, the ambient light characteristic data is only meaningful during periods of sufficient natural light. Therefore, the target device's time can be used to determine whether the ambient light characteristic data should be used as input data for the state recognition model to initially identify the target device as being in the set state at time i. The same principle applies when training the state recognition model: The speed characteristic data can be used to determine whether the magnetic field characteristic data should be used as training input data for the state recognition model, and the device's time can be used to determine whether the ambient light characteristic data should be used as training input data for the state recognition model.

[0143] That is, in the present disclosure, the first training module is further configured as follows:

[0144] If the speed characteristic data is lower than a preset speed threshold, the sensor characteristic data further includes: magnetic field characteristic data;

[0145] If it is determined that the device is in a time period with sufficient lighting based on the time of the device corresponding to the sensor characteristic data, the sensor characteristic data further includes: ambient light characteristic data.

[0146] As mentioned above, to further improve state recognition accuracy, the sensor feature data input into the state recognition model for preliminary identification of the target device's set state at time i may also include map feature data for the target device. When the sensor feature data input into the state recognition model also includes map feature data, when training the state recognition model, the sensor feature data used as training input for the state recognition model must also include map feature data for the device corresponding to the sensor feature data.

[0147] In the present disclosure, the apparatus may further include:

[0148] The second training module obtains training data for the state transition model, where the training data includes sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data, ambient light feature data, and network signal feature data; from the sensor feature data, respectively obtains signal strength feature data representing changes in satellite signal strength, ambient light intensity, and network signal strength; determines the signal strength change label corresponding to the device and the signal strength feature data; uses the signal strength feature data and the signal strength change label corresponding to the signal strength feature data as training input data for the state transition recognition model to train the state transition recognition model.

[0149] In this embodiment, when training the state transition recognition model, training data of the state transition model including sensor feature data of a device is first obtained, wherein the sensor feature data includes at least satellite positioning feature data, ambient light feature data, and network signal feature data; then, signal strength feature data for characterizing changes in satellite signal strength, ambient light intensity, and network signal strength are respectively obtained from the sensor feature data; then, a signal strength change label corresponding to the device and the signal strength feature data is determined. For example, if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the device and the signal strength feature data is "significant satellite signal enhancement"; if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is significantly weakened, then the signal strength change label corresponding to the device and the signal strength feature data is "significant satellite signal weakening"; if it is determined according to the signal strength feature data that the satellite signal received by the corresponding device is unchanged, then the signal strength change label corresponding to the device and the signal strength feature data is "satellite signal unchanged", and so on; if it is determined according to the signal strength feature data that the ambient light detected by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the device and the signal strength feature data is "ambient light enhancement" Obviously, if it is determined according to the signal strength characteristic data that the ambient light detected by the corresponding device is significantly weakened, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the ambient light is significantly weakened; if it is determined according to the signal strength characteristic data that the ambient light detected by the corresponding device is unchanged, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the ambient light is unchanged, and so on; if it is determined according to the signal strength characteristic data that the network signal received by the corresponding device is significantly enhanced, then the signal strength change label corresponding to the signal strength characteristic data of the device is that the network signal is significantly enhanced; if it is determined according to the signal strength characteristic data that the network signal received by the corresponding device is significantly enhanced If the network signal is significantly weakened, the signal strength change label corresponding to the device and the signal strength characteristic data is that the network signal is significantly weakened. If it is determined based on the signal strength characteristic data that the network signal received by the corresponding device is unchanged, then the signal strength change label corresponding to the device and the signal strength characteristic data is that the network signal is unchanged, and so on. Finally, the signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data are used as training input data for the state transition recognition model to train the state transition recognition model. When the state transition recognition model meets the preset convergence condition, the training can be terminated to obtain the state transition recognition model.

[0150] In the present disclosure, in order to further improve the recognition accuracy of the state transition recognition model, further sliding window processing may be performed on the signal strength feature data.

[0151] That is, in the present disclosure, the second training module may also be configured as:

[0152] Based on the time of the signal strength characteristic data, according to a preset time length, counting the characteristic data extreme values ​​of the signal strength characteristic data in different time sliding windows;

[0153] Calculating a characteristic data difference between the characteristic data extreme value and the value of the signal strength characteristic data at the current moment;

[0154] The using the signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data as training input data of the state transition recognition model to train the state transition recognition model includes:

[0155] The characteristic data difference and the signal strength change label corresponding to the characteristic data difference are used as training input data of the state transition recognition model to train the state transition recognition model.

[0156] In this embodiment, first, based on the time of the signal strength characteristic data, according to the preset time length, the characteristic data extreme values ​​of the signal strength characteristic data in different time sliding windows are counted, wherein the preset time length can be set according to the needs of the actual application, for example, it can be set to 5s or other values, wherein for visible satellites, the maximum value of the corresponding characteristic data can be counted, and for theoretically visible satellites, the minimum value of the corresponding characteristic data can be counted; then, the characteristic data difference between the statistically obtained characteristic data extreme value and the value of the signal strength characteristic data at the current moment is calculated; finally, the characteristic data difference is used to replace the signal strength characteristic data received by the training device to train the state transition recognition model, that is, the characteristic data difference and the signal strength change label corresponding to the characteristic data difference are used as the training input data of the state transition recognition model, and the state transition recognition model is trained. When the state transition recognition model meets the preset convergence condition, the training can be ended to obtain the state transition recognition model.

[0157] The present disclosure also discloses an electronic device, Figure 4 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown. Figure 4 As shown, the electronic device 400 includes a memory 401 and a processor 402; wherein,

[0158] The memory 401 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 402 to implement the above method steps.

[0159] Figure 5 It is a structural diagram of a computer system suitable for implementing the device status identification method according to an embodiment of the present disclosure.

[0160] like Figure 5 As shown, the computer system 500 includes a processing unit 501, which can execute various processes in the above-mentioned embodiments according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage unit 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the computer system 500 are also stored in the RAM 503. The processing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0161] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed so that a computer program read therefrom can be installed into the storage section 508 as needed. Among them, the processing unit 501 can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, an NPU, etc.

[0162] In particular, according to embodiments of the present disclosure, the methods described above can be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program tangibly embodied on a computer-readable medium, the computer program comprising program code for executing the methods described. In such embodiments, the computer program can be downloaded and installed from a network via the communication portion 509 and / or installed from the removable medium 511.

[0163] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the diagram or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, as well as the combination of boxes in the block diagram and / or flow chart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0164] The units or modules described in the embodiments of the present disclosure may be implemented in software or hardware. The units or modules described may also be provided in a processor, and the names of these units or modules do not, in certain circumstances, limit the units or modules themselves.

[0165] As another aspect, the present disclosure further provides a computer-readable storage medium. This computer-readable storage medium may be included in the apparatus described in the above embodiments, or may be a standalone computer-readable storage medium not incorporated into the apparatus. The computer-readable storage medium stores one or more programs, which are used by one or more processors to execute the methods described in the present disclosure.

[0166] The above description is merely a preferred embodiment of the present disclosure and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention herein is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this disclosure.

Claims

1. A device status identification method, wherein: include: Acquire sensor feature data of the target device at the i-th moment, wherein the sensor feature data at least includes satellite positioning feature data; Inputting the sensor feature data at the i-th moment into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where there are at least two set states; Inputting the initial confidence that the target device is in the set state at the i-th moment and the historical revised confidence that the target device is in the set state before the i-th moment into a pre-trained state smoothing model to obtain the revised confidence that the target device is in the set state at the i-th moment; According to the modified confidence that the target device is in the set state at the i-th moment, the actual state of the target device is determined from the set state.

2. The method according to claim 1, wherein The sensor characteristic data further includes: speed characteristic data, magnetic field characteristic data and ambient light characteristic data, and the method further includes: Based on the speed characteristic data, determining whether the speed of the target device is lower than a preset speed threshold, and if lower than the speed threshold, the sensor characteristic data input into the pre-trained state recognition model also includes: magnetic field characteristic data; Based on the time of the target device, it is determined whether the target device is in a time period with sufficient lighting. If so, the sensor feature data input into the pre-trained state recognition model also includes: ambient light feature data.

3. The method according to claim 2, wherein: The sensor characteristic data includes: network signal characteristic data, and the method further includes: Obtaining signal strength characteristic data representing changes in satellite signal strength, ambient light intensity, and network signal strength from the satellite positioning characteristic data, ambient light characteristic data, and network signal characteristic data included in the sensor characteristic data acquired at and before the i-th moment; Inputting the signal strength characteristic data into a pre-trained state transition recognition model to obtain a signal strength change recognition result of the target device; The signal strength change recognition result is input into the state smoothing model, and the value of the state transition probability of the state smoothing model is adjusted. The value of the state transition probability is used to correct the initial confidence that the target device is in the set state at the i-th moment.

4. The method according to claim 3, wherein: The set state includes: indoor, outdoor, and semi-indoor, and the adjusting of the value of the state transition probability of the state smoothing model includes: When the signal strength change identification result at least indicates that the satellite signal received by the target device is significantly enhanced or the ambient light is significantly enhanced, the state transition probability values ​​from indoor to outdoor and from semi-indoor to outdoor in the state smoothing model are increased respectively; When the signal strength change identification result at least indicates that the network signal received by the target device is significantly enhanced, the satellite signal is significantly weakened, or the ambient light is significantly weakened, the state transition probability values ​​from outdoor to indoor and from semi-indoor to indoor in the state smoothing model are increased respectively.

5. The method according to any one of claims 2 to 4, wherein: Also includes: Acquire training data for a state recognition model, the training data including sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data; Determine a status tag of a device corresponding to the sensor characteristic data; The sensor feature data and the state label of the corresponding device are used as training input data of the state recognition model to train the state recognition model.

6. The method according to claim 5, wherein: The sensor characteristic data further includes speed characteristic data and ambient light characteristic data, and the magnetic field characteristic data is used for the state recognition model learning device to learn a state in which the speed characteristic data is lower than a preset speed threshold; The ambient light feature data is used to determine whether the state recognition model learning device is in a time period with sufficient light.

7. The method according to claim 3 or 4, wherein: Also includes: Acquire training data for a state transition model, the training data including sensor feature data of the device, wherein the sensor feature data includes at least satellite positioning feature data, ambient light feature data, and network signal feature data; Acquire signal strength characteristic data representing changes in satellite signal strength, changes in ambient light intensity, and changes in network signal strength from the sensor characteristic data; determining a signal strength change tag corresponding to the device and the signal strength characteristic data; The signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data are used as training input data of the state transition recognition model to train the state transition recognition model.

8. The method according to claim 7, wherein: Also includes: Based on the time of the signal strength characteristic data, according to a preset time length, counting the characteristic data extreme values ​​of the signal strength characteristic data in different time sliding windows; Calculating a characteristic data difference between the characteristic data extreme value and the value of the signal strength characteristic data at the current moment; The using the signal strength characteristic data and the signal strength change label corresponding to the signal strength characteristic data as training input data of the state transition recognition model to train the state transition recognition model includes: The characteristic data difference and the signal strength change label corresponding to the characteristic data difference are used as training input data of the state transition recognition model to train the state transition recognition model.

9. A device for identifying device status, wherein: The device comprises: an acquisition module, which acquires sensor feature data of the target device at the i-th moment, wherein the sensor feature data at least includes satellite positioning feature data; a pre-identification module, inputting the sensor feature data at the i-th moment into a pre-trained state recognition model to obtain an initial confidence level that the target device is in a set state at the i-th moment, where the set states are at least two; a correction module, inputting the initial confidence that the target device is in the set state at the i-th moment and the historical corrected confidence that the target device is in the set state before the i-th moment into a pre-trained state smoothing model to obtain a corrected confidence that the target device is in the set state at the i-th moment; The determination module determines the actual state of the target device from the set state according to the modified confidence that the target device is in the set state at the i-th moment.

10. An electronic device comprising a memory and at least one processor; wherein: The memory is configured to store one or more computer instructions, wherein the one or more computer instructions are executed by the at least one processor to implement the method according to any one of claims 1 to 8.

Images